>  Docs Center  >  Libraries  >  Markwardt  >  MPFITEXPR






  Craig B. Markwardt, NASA/GSFC Code 662, Greenbelt, MD 20770
  UPDATED VERSIONs can be found on my WEB PAGE:


  Perform Levenberg-Marquardt least-squares fit to arbitrary expression

Major Topics

  Curve and Surface Fitting

Calling Sequence

  MYFUNCT = 'X*(1-X)+3'
  parms = MPFITEXPR(MYFUNCT, XVAL, YVAL, ERR, start_parms, ...)


  MPFITEXPR fits a user-supplied model -- in the form of an arbitrary IDL
  expression -- to a set of user-supplied data. MPFITEXPR calls
  MPFIT, the MINPACK-1 least-squares minimizer, to do the main
  Given the data and their uncertainties, MPFITEXPR finds the best set
  of model parameters which match the data (in a least-squares
  sense) and returns them in an array.
  The user must supply the following items:
  - An array of independent variable values ("X").
  - An array of "measured" *dependent* variable values ("Y").
  - An array of "measured" 1-sigma uncertainty values ("ERR").
  - A text IDL expression which computes Y given X.
  - Starting guesses for all of the parameters ("START_PARAMS").
  There are very few restrictions placed on X, Y or the expression of
  the model. Simply put, the expression must map the "X" values into
  "Y" values given the model parameters. The "X" values may
  represent any independent variable (not just Cartesian X), and
  indeed may be multidimensional themselves. For example, in the
  application of image fitting, X may be a 2xN array of image
  Some rules must be obeyed in constructing the expression. First,
  the independent variable name *MUST* be "X" in the expression,
  regardless of the name of the variable being passed to MPFITEXPR.
  This is demonstrated in the above calling sequence, where the X
  variable passed in is called "XVAL" but the expression still refers
  to "X". Second, parameter values must be referred to as an array
  named "P".
  If you do not pass in starting values for the model parameters,
  MPFITEXPR will attempt to determine the number of parameters you
  intend to have (it does this by looking for references to the array
  variable named "P"). When no starting values are passed in, the
  values are assumed to start at zero.
  MPFITEXPR carefully avoids passing large arrays where possible to
  improve performance.
  See below for an example of usage.
  This source module also provides a function called MPEVALEXPR. You
  can use this function to evaluate your expression, given a list of
  parameters. This is one of the easier ways to compute the model
  once the best-fit parameters have been found. Here is an example:
  where MYFUNCT is the expression (see MYFUNCT below), XVAL is the
  list of "X" values, and PARMS is an array of parameters. The
  returned array YMOD contains the expression MYFUNCT evaluated at
  each point in XVAL.
  The most complicated optimization problems typically involve other
  external parameters, in addition to the fitted parameters. While
  it is extremely easy to rewrite an expression dynamically, in case
  one of the external parameters changes, the other possibility is to
  use the PRIVATE data structure.
  The user merely passes a structure to the FUNCTARGS keyword. The
  user expression receives this value as the variable PRIVATE.
  MPFITEXPR nevers accesses this variable so it can contain any
  desired values. Usually it would be an IDL structure so that any
  named external parameters can be passed to the expression.

Constraining Parameter Values With The Parinfo Keyword

  The behavior of MPFIT can be modified with respect to each
  parameter to be fitted. A parameter value can be fixed; simple
  boundary constraints can be imposed; limitations on the parameter
  changes can be imposed; properties of the automatic derivative can
  be modified; and parameters can be tied to one another.
  These properties are governed by the PARINFO structure, which is
  passed as a keyword parameter to MPFIT.
  PARINFO should be an array of structures, one for each parameter.
  Each parameter is associated with one element of the array, in
  numerical order. The structure can have the following entries
  (none are required):
    .VALUE - the starting parameter value (but see the START_PARAMS
              parameter for more information).
    .FIXED - a boolean value, whether the parameter is to be held
              fixed or not. Fixed parameters are not varied by
              MPFIT, but are passed on to MYFUNCT for evaluation.
    .LIMITED - a two-element boolean array. If the first/second
                element is set, then the parameter is bounded on the
                lower/upper side. A parameter can be bounded on both
                sides. Both LIMITED and LIMITS must be given
    .LIMITS - a two-element float or double array. Gives the
              parameter limits on the lower and upper sides,
              respectively. Zero, one or two of these values can be
              set, depending on the values of LIMITED. Both LIMITED
              and LIMITS must be given together.
    .PARNAME - a string, giving the name of the parameter. The
                fitting code of MPFIT does not use this tag in any
                way. However, the default ITERPROC will print the
                parameter name if available.
    .STEP - the step size to be used in calculating the numerical
            derivatives. If set to zero, then the step size is
            computed automatically. Ignored when AUTODERIVATIVE=0.
            This value is superceded by the RELSTEP value.
    .RELSTEP - the *relative* step size to be used in calculating
                the numerical derivatives. This number is the
                fractional size of the step, compared to the
                parameter value. This value supercedes the STEP
                setting. If the parameter is zero, then a default
                step size is chosen.
    .MPSIDE - the sidedness of the finite difference when computing
              numerical derivatives. This field can take four
                  0 - one-sided derivative computed automatically
                  1 - one-sided derivative (f(x+h) - f(x) )/h
                -1 - one-sided derivative (f(x) - f(x-h))/h
                  2 - two-sided derivative (f(x+h) - f(x-h))/(2*h)
              Where H is the STEP parameter described above. The
              "automatic" one-sided derivative method will chose a
              direction for the finite difference which does not
              violate any constraints. The other methods do not
              perform this check. The two-sided method is in
              principle more precise, but requires twice as many
              function evaluations. Default: 0.
    .MPMAXSTEP - the maximum change to be made in the parameter
                  value. During the fitting process, the parameter
                  will never be changed by more than this value in
                  one iteration.
                  A value of 0 indicates no maximum. Default: 0.
    .TIED - a string expression which "ties" the parameter to other
            free or fixed parameters as an equality constraint. Any
            expression involving constants and the parameter array P
            are permitted.
            Example: if parameter 2 is always to be twice parameter
            1 then use the following: parinfo[2].tied = '2 * P[1]'.
            Since they are totally constrained, tied parameters are
            considered to be fixed; no errors are computed for them.
            [ NOTE: the PARNAME can't be used in a TIED expression. ]
    .MPPRINT - if set to 1, then the default ITERPROC will print the
                parameter value. If set to 0, the parameter value
                will not be printed. This tag can be used to
                selectively print only a few parameter values out of
                many. Default: 1 (all parameters printed)
    .MPFORMAT - IDL format string to print the parameter within
                ITERPROC. Default: '(G20.6)' (An empty string will
                also use the default.)
  Future modifications to the PARINFO structure, if any, will involve
  adding structure tags beginning with the two letters "MP".
  Therefore programmers are urged to avoid using tags starting with
  "MP", but otherwise they are free to include their own fields
  within the PARINFO structure, which will be ignored by MPFIT.
  PARINFO Example:
  parinfo = replicate({value:0.D, fixed:0, limited:[0,0], $
                      limits:[0.D,0]}, 5)
  parinfo[0].fixed = 1
  parinfo[4].limited[0] = 1
  parinfo[4].limits[0] = 50.D
  parinfo[*].value = [5.7D, 2.2, 500., 1.5, 2000.]
  A total of 5 parameters, with starting values of 5.7,
  2.2, 500, 1.5, and 2000 are given. The first parameter
  is fixed at a value of 5.7, and the last parameter is
  constrained to be above 50.


  This function is designed to work with IDL 5.0 or greater. Because
  this function uses the IDL EXECUTE() function, it will not work
  with the free version of the IDL Virtual machine.


  MYFUNCT - a string variable containing an IDL expression. The
            only restriction is that the independent variable *must*
            be referred to as "X" and model parameters *must* be
            referred to as an array called "P". Do not use symbol
            names beginning with the underscore, "_".
            The expression should calculate "model" Y values given
            the X values and model parameters. Using the vector
            notation of IDL, this can be quite easy to do. If your
            expression gets complicated, you may wish to make an IDL
            function which will improve performance and readibility.
            The resulting array should be of the same size and
            dimensions as the "measured" Y values.
  X - Array of independent variable values.
  Y - Array of "measured" dependent variable values. Y should have
      the same data type as X. The function MYFUNCT should map
  ERR - Array of "measured" 1-sigma uncertainties. ERR should have
        the same data type as Y. ERR is ignored if the WEIGHTS
        keyword is specified.
  START_PARAMS - An array of starting values for each of the
                  parameters of the model. The number of parameters
                  should be fewer than the number of measurements.
                  Also, the parameters should have the same data type
                  as the measurements (double is preferred).
                  This parameter is optional if the PARINFO keyword
                  is used (see MPFIT). The PARINFO keyword provides
                  a mechanism to fix or constrain individual
                  parameters. If both START_PARAMS and PARINFO are
                  passed, then the starting *value* is taken from
                  START_PARAMS, but the *constraints* are taken from
                  If no parameters are given, then MPFITEXPR attempts
                  to determine the number of parameters by scanning
                  the expression. Parameters determined this way are
                  initialized to zero. This technique is not fully
                  reliable, so users are advised to pass explicit
                  parameter starting values.


  Returns the array of best-fit parameters.

Keyword Parameters

  BESTNORM - the value of the summed, squared, weighted residuals
              for the returned parameter values, i.e. the chi-square value.
  COVAR - the covariance matrix for the set of parameters returned
          by MPFIT. The matrix is NxN where N is the number of
          parameters. The square root of the diagonal elements
          gives the formal 1-sigma statistical errors on the
          parameters IF errors were treated "properly" in MYFUNC.
          Parameter errors are also returned in PERROR.
          To compute the correlation matrix, PCOR, use this:
          IDL> PCOR = COV * 0
          IDL> FOR i = 0, n-1 DO FOR j = 0, n-1 DO $
                PCOR[i,j] = COV[i,j]/sqrt(COV[i,i]*COV[j,j])
          If NOCOVAR is set or MPFIT terminated abnormally, then
          COVAR is set to a scalar with value !VALUES.D_NAN.
  DOF - number of degrees of freedom, computed as
        Note that this doesn't account for pegged parameters (see
  ERRMSG - a string error or warning message is returned.
  FTOL - a nonnegative input variable. Termination occurs when both
          the actual and predicted relative reductions in the sum of
          squares are at most FTOL (and STATUS is accordingly set to
          1 or 3). Therefore, FTOL measures the relative error
          desired in the sum of squares. Default: 1D-10
  FUNCTARGS - passed directly to the expression as the variable
              PRIVATE. Any user-private data can be communicated to
              the user expression using this keyword.
              Default: PRIVATE is undefined in user expression
  GTOL - a nonnegative input variable. Termination occurs when the
          cosine of the angle between fvec and any column of the
          jacobian is at most GTOL in absolute value (and STATUS is
          accordingly set to 4). Therefore, GTOL measures the
          orthogonality desired between the function vector and the
          columns of the jacobian. Default: 1D-10
  ITERARGS - The keyword arguments to be passed to ITERPROC via the
              _EXTRA mechanism. This should be a structure, and is
              similar in operation to FUNCTARGS.
              Default: no arguments are passed.
  ITERPROC - The name of a procedure to be called upon each NPRINT
              iteration of the MPFIT routine. It should be declared
              in the following way:
              PRO ITERPROC, MYFUNCT, p, iter, fnorm, FUNCTARGS=fcnargs, $
                PARINFO=parinfo, QUIET=quiet, ...
                ; perform custom iteration update
              ITERPROC must either accept all three keyword
              parameters (FUNCTARGS, PARINFO and QUIET), or at least
              accept them via the _EXTRA keyword.
              MYFUNCT is the user-supplied function to be minimized,
              P is the current set of model parameters, ITER is the
              iteration number, and FUNCTARGS are the arguments to be
              passed to MYFUNCT. FNORM should be the
              chi-squared value. QUIET is set when no textual output
              should be printed. See below for documentation of
              In implementation, ITERPROC can perform updates to the
              terminal or graphical user interface, to provide
              feedback while the fit proceeds. If the fit is to be
              stopped for any reason, then ITERPROC should set the
              common block variable ERROR_CODE to negative value (see
              MPFIT_ERROR common block below). In principle,
              ITERPROC should probably not modify the parameter
              values, because it may interfere with the algorithm's
              stability. In practice it is allowed.
              Default: an internal routine is used to print the
                      parameter values.
  MAXITER - The maximum number of iterations to perform. If the
            number is exceeded, then the STATUS value is set to 5
            and MPFIT returns.
            Default: 200 iterations
  NFEV - the number of MYFUNCT function evaluations performed.
  NFREE - the number of free parameters in the fit. This includes
          parameters which are not FIXED and not TIED, but it does
          include parameters which are pegged at LIMITS.
  NITER - the number of iterations completed.
  NOCOVAR - set this keyword to prevent the calculation of the
            covariance matrix before returning (see COVAR)
  NPEGGED - the number of free parameters which are pegged at a
  NPRINT - The frequency with which ITERPROC is called. A value of
            1 indicates that ITERPROC is called with every iteration,
            while 2 indicates every other iteration, etc. Note that
            several Levenberg-Marquardt attempts can be made in a
            single iteration.
            Default value: 1
  PARINFO - Provides a mechanism for more sophisticated constraints
            to be placed on parameter values. When PARINFO is not
            passed, then it is assumed that all parameters are free
            and unconstrained. Values in PARINFO are never
            modified during a call to MPFIT.
            See description above for the structure of PARINFO.
            Default value: all parameters are free and unconstrained.
  PERROR - The formal 1-sigma errors in each parameter, computed
            from the covariance matrix. If a parameter is held
            fixed, or if it touches a boundary, then the error is
            reported as zero.
            If the fit is unweighted (i.e. no errors were given, or
            the weights were uniformly set to unity), then PERROR
            will probably not represent the true parameter
            *If* you can assume that the true reduced chi-squared
            value is unity -- meaning that the fit is implicitly
            assumed to be of good quality -- then the estimated
            parameter uncertainties can be computed by scaling PERROR
            by the measured chi-squared value.
              DOF = N_ELEMENTS(X) - N_ELEMENTS(PARMS) ; deg of freedom
              PCERROR = PERROR * SQRT(BESTNORM / DOF) ; scaled uncertainties
  QUIET - set this keyword when no textual output should be printed
          by MPFIT
  STATUS - an integer status code is returned. All values other
            than zero can represent success. It can have one of the
            following values:
0 improper input parameters.
1 both actual and predicted relative reductions
in the sum of squares are at most FTOL.
2 relative error between two consecutive iterates
is at most XTOL
3 conditions for STATUS = 1 and STATUS = 2 both hold.
4 the cosine of the angle between fvec and any
column of the jacobian is at most GTOL in
absolute value.
5 the maximum number of iterations has been reached
6 FTOL is too small. no further reduction in
the sum of squares is possible.
7 XTOL is too small. no further improvement in
the approximate solution x is possible.
8 GTOL is too small. fvec is orthogonal to the
columns of the jacobian to machine precision.
  WEIGHTS - Array of weights to be used in calculating the
            chi-squared value. If WEIGHTS is specified then the ERR
            parameter is ignored. The chi-squared value is computed
            as follows:
                CHISQ = TOTAL( (Y-MYFUNCT(X,P))^2 * ABS(WEIGHTS) )
            Here are common values of WEIGHTS:
                1D/ERR^2 - Normal weighting (ERR is the measurement error)
                1D/Y - Poisson weighting (counting statistics)
                1D - Unweighted
  XTOL - a nonnegative input variable. Termination occurs when the
          relative error between two consecutive iterates is at most
          XTOL (and STATUS is accordingly set to 2 or 3). Therefore,
          XTOL measures the relative error desired in the approximate
          solution. Default: 1D-10
  YFIT - the best-fit model function, as returned by MYFUNCT.


  ; First, generate some synthetic data
  x = dindgen(200) * 0.1 - 10. ; Independent variable
  yi = gauss1(x, [2.2D, 1.4, 3000.]) + 1000 ; "Ideal" Y variable
  y = yi + randomn(seed, 200) * sqrt(yi) ; Measured, w/ noise
  sy = sqrt(y) ; Poisson errors
  ; Now fit a Gaussian to see how well we can recover
  expr = 'P[0] + GAUSS1(X, P[1:3])' ; fitting function
  p0 = [800.D, 1.D, 1., 500.] ; Initial guess
  p = mpfitexpr(expr, x, y, sy, p0) ; Fit the expression
  print, p
  plot, x, y ; Plot data
  oplot, x, mpevalexpr(expr, x, p) ; Plot model
  Generates a synthetic data set with a Gaussian peak, and Poisson
  statistical uncertainty. Then a model consisting of a constant
  plus Gaussian is fit to the data.

Common Blocks

    User routines may stop the fitting process at any time by
    setting an error condition. This condition may be set in either
    the user's model computation routine (MYFUNCT), or in the
    iteration procedure (ITERPROC).
    To stop the fitting, the above common block must be declared,
    and ERROR_CODE must be set to a negative number. After the user
    procedure or function returns, MPFIT checks the value of this
    common block variable and exits immediately if the error
    condition has been set. By default the value of ERROR_CODE is
    zero, indicating a successful function/procedure call.


  MINPACK-1, Jorge More', available from netlib (www.netlib.org).
  "Optimization Software Guide," Jorge More' and Stephen Wright,
    SIAM, *Frontiers in Applied Mathematics*, Number 14.

Modification History

  Written, Apr-Jul 1998, CM
  Added PERROR keyword, 04 Aug 1998, CM
  Added COVAR keyword, 20 Aug 1998, CM
  Added NITER output keyword, 05 Oct 1998
      D.L Windt, Bell Labs, windt@bell-labs.com;
  Added ability to return model function in YFIT, 09 Nov 1998
  Parameter values can be tied to others, 09 Nov 1998
  Added MPEVALEXPR utility function, 09 Dec 1998
  Cosmetic documentation updates, 16 Apr 1999, CM
  More cosmetic documentation updates, 14 May 1999, CM
  Made sure to update STATUS value, 25 Sep 1999, CM
  Added WEIGHTS keyword, 25 Sep 1999, CM
  Changed from handles to common blocks, 25 Sep 1999, CM
    - commons seem much cleaner and more logical in this case.
  Alphabetized documented keywords, 02 Oct 1999, CM
  Added QUERY keyword and query checking of MPFIT, 29 Oct 1999, CM
  Check to be sure that X and Y are present, 02 Nov 1999, CM
  Documented PERROR for unweighted fits, 03 Nov 1999, CM
  Removed ITERPROC='' when quiet EQ 1, 10 Jan 2000, CM
  Changed to ERROR_CODE for error condition, 28 Jan 2000, CM
  Updated the EXAMPLE, 26 Mar 2000, CM
  Copying permission terms have been liberalized, 26 Mar 2000, CM
  Propagated improvements from MPFIT, 17 Dec 2000, CM
  Correct reference to _WTS in MPFITEXPR_EVAL, 25 May 2001, CM
      (thanks to Mark Fardal)
  Added useful FUNCTARGS behavior (as yet undocumented), 04 Jul
      2002, CM
  Documented FUNCTARGS/PRIVATE behavior, 22 Jul 2002, CM
  Added NFREE and NPEGGED keywords, 13 Sep 2002, CM
  Documented RELSTEP field of PARINFO (!!), CM, 25 Oct 2002
  Add DOF keyword, CM, 31 Jul 2003
  Add FUNCTARGS keyword to MPEVALEXPR, CM 08 Aug 2003
  Minor documentation adjustment, 03 Feb 2004, CM
  Move STRICTARR compile option inside each function/procedure, 9 Oct 2006
  Clarify documentation on user-function, derivatives, and PARINFO,
    27 May 2007
  Add COMPATIBILITY section, CM, 13 Dec 2007

© 2020 Harris Geospatial Solutions, Inc. |  Legal
My Account    |    Store    |    Contact Us